]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - kernel/sysctl.c
projects / mirror_ubuntu-artful-kernel.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
mm: per device dirty threshold
[mirror_ubuntu-artful-kernel.git] / kernel / sysctl.c
1 /*
2 * sysctl.c: General linux system control interface
3 *
4 * Begun 24 March 1995, Stephen Tweedie
5 * Added /proc support, Dec 1995
6 * Added bdflush entry and intvec min/max checking, 2/23/96, Tom Dyas.
7 * Added hooks for /proc/sys/net (minor, minor patch), 96/4/1, Mike Shaver.
8 * Added kernel/java-{interpreter,appletviewer}, 96/5/10, Mike Shaver.
9 * Dynamic registration fixes, Stephen Tweedie.
10 * Added kswapd-interval, ctrl-alt-del, printk stuff, 1/8/97, Chris Horn.
11 * Made sysctl support optional via CONFIG_SYSCTL, 1/10/97, Chris
12 * Horn.
13 * Added proc_doulongvec_ms_jiffies_minmax, 09/08/99, Carlos H. Bauer.
14 * Added proc_doulongvec_minmax, 09/08/99, Carlos H. Bauer.
15 * Changed linked lists to use list.h instead of lists.h, 02/24/00, Bill
16 * Wendling.
17 * The list_for_each() macro wasn't appropriate for the sysctl loop.
18 * Removed it and replaced it with older style, 03/23/00, Bill Wendling
19 */
20
21 #include <linux/module.h>
22 #include <linux/mm.h>
23 #include <linux/swap.h>
24 #include <linux/slab.h>
25 #include <linux/sysctl.h>
26 #include <linux/proc_fs.h>
27 #include <linux/capability.h>
28 #include <linux/ctype.h>
29 #include <linux/utsname.h>
30 #include <linux/smp_lock.h>
31 #include <linux/fs.h>
32 #include <linux/init.h>
33 #include <linux/kernel.h>
34 #include <linux/kobject.h>
35 #include <linux/net.h>
36 #include <linux/sysrq.h>
37 #include <linux/highuid.h>
38 #include <linux/writeback.h>
39 #include <linux/hugetlb.h>
40 #include <linux/security.h>
41 #include <linux/initrd.h>
42 #include <linux/times.h>
43 #include <linux/limits.h>
44 #include <linux/dcache.h>
45 #include <linux/syscalls.h>
46 #include <linux/nfs_fs.h>
47 #include <linux/acpi.h>
48 #include <linux/reboot.h>
49
50 #include <asm/uaccess.h>
51 #include <asm/processor.h>
52
53 #ifdef CONFIG_X86
54 #include <asm/nmi.h>
55 #include <asm/stacktrace.h>
56 #endif
57
58 #if defined(CONFIG_SYSCTL)
59
60 /* External variables not in a header file. */
61 extern int C_A_D;
62 extern int print_fatal_signals;
63 extern int sysctl_overcommit_memory;
64 extern int sysctl_overcommit_ratio;
65 extern int sysctl_panic_on_oom;
66 extern int max_threads;
67 extern int core_uses_pid;
68 extern int suid_dumpable;
69 extern char core_pattern[];
70 extern int pid_max;
71 extern int min_free_kbytes;
72 extern int printk_ratelimit_jiffies;
73 extern int printk_ratelimit_burst;
74 extern int pid_max_min, pid_max_max;
75 extern int sysctl_drop_caches;
76 extern int percpu_pagelist_fraction;
77 extern int compat_log;
78 extern int maps_protect;
79 extern int sysctl_stat_interval;
80 extern int audit_argv_kb;
81
82 /* this is needed for the proc_dointvec_minmax for [fs_]overflow UID and GID */
83 static int maxolduid = 65535;
84 static int minolduid;
85 static int min_percpu_pagelist_fract = 8;
86
87 static int ngroups_max = NGROUPS_MAX;
88
89 #ifdef CONFIG_KMOD
90 extern char modprobe_path[];
91 #endif
92 #ifdef CONFIG_CHR_DEV_SG
93 extern int sg_big_buff;
94 #endif
95
96 #ifdef __sparc__
97 extern char reboot_command [];
98 extern int stop_a_enabled;
99 extern int scons_pwroff;
100 #endif
101
102 #ifdef __hppa__
103 extern int pwrsw_enabled;
104 extern int unaligned_enabled;
105 #endif
106
107 #ifdef CONFIG_S390
108 #ifdef CONFIG_MATHEMU
109 extern int sysctl_ieee_emulation_warnings;
110 #endif
111 extern int sysctl_userprocess_debug;
112 extern int spin_retry;
113 #endif
114
115 extern int sysctl_hz_timer;
116
117 #ifdef CONFIG_BSD_PROCESS_ACCT
118 extern int acct_parm[];
119 #endif
120
121 #ifdef CONFIG_IA64
122 extern int no_unaligned_warning;
123 #endif
124
125 #ifdef CONFIG_RT_MUTEXES
126 extern int max_lock_depth;
127 #endif
128
129 #ifdef CONFIG_SYSCTL_SYSCALL
130 static int parse_table(int __user *, int, void __user *, size_t __user *,
131 void __user *, size_t, ctl_table *);
132 #endif
133
134
135 #ifdef CONFIG_PROC_SYSCTL
136 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
137 void __user *buffer, size_t *lenp, loff_t *ppos);
138 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
139 void __user *buffer, size_t *lenp, loff_t *ppos);
140 #endif
141
142 static ctl_table root_table[];
143 static struct ctl_table_header root_table_header =
144 { root_table, LIST_HEAD_INIT(root_table_header.ctl_entry) };
145
146 static ctl_table kern_table[];
147 static ctl_table vm_table[];
148 static ctl_table fs_table[];
149 static ctl_table debug_table[];
150 static ctl_table dev_table[];
151 extern ctl_table random_table[];
152 #ifdef CONFIG_UNIX98_PTYS
153 extern ctl_table pty_table[];
154 #endif
155 #ifdef CONFIG_INOTIFY_USER
156 extern ctl_table inotify_table[];
157 #endif
158
159 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
160 int sysctl_legacy_va_layout;
161 #endif
162
163 extern int prove_locking;
164 extern int lock_stat;
165
166 /* The default sysctl tables: */
167
168 static ctl_table root_table[] = {
169 {
170 .ctl_name = CTL_KERN,
171 .procname = "kernel",
172 .mode = 0555,
173 .child = kern_table,
174 },
175 {
176 .ctl_name = CTL_VM,
177 .procname = "vm",
178 .mode = 0555,
179 .child = vm_table,
180 },
181 #ifdef CONFIG_NET
182 {
183 .ctl_name = CTL_NET,
184 .procname = "net",
185 .mode = 0555,
186 .child = net_table,
187 },
188 #endif
189 {
190 .ctl_name = CTL_FS,
191 .procname = "fs",
192 .mode = 0555,
193 .child = fs_table,
194 },
195 {
196 .ctl_name = CTL_DEBUG,
197 .procname = "debug",
198 .mode = 0555,
199 .child = debug_table,
200 },
201 {
202 .ctl_name = CTL_DEV,
203 .procname = "dev",
204 .mode = 0555,
205 .child = dev_table,
206 },
207 /*
208 * NOTE: do not add new entries to this table unless you have read
209 * Documentation/sysctl/ctl_unnumbered.txt
210 */
211 { .ctl_name = 0 }
212 };
213
214 #ifdef CONFIG_SCHED_DEBUG
215 static unsigned long min_sched_granularity_ns = 100000; /* 100 usecs */
216 static unsigned long max_sched_granularity_ns = 1000000000; /* 1 second */
217 static unsigned long min_wakeup_granularity_ns; /* 0 usecs */
218 static unsigned long max_wakeup_granularity_ns = 1000000000; /* 1 second */
219 #endif
220
221 static ctl_table kern_table[] = {
222 #ifdef CONFIG_SCHED_DEBUG
223 {
224 .ctl_name = CTL_UNNUMBERED,
225 .procname = "sched_nr_latency",
226 .data = &sysctl_sched_nr_latency,
227 .maxlen = sizeof(unsigned int),
228 .mode = 0644,
229 .proc_handler = &proc_dointvec,
230 },
231 {
232 .ctl_name = CTL_UNNUMBERED,
233 .procname = "sched_latency_ns",
234 .data = &sysctl_sched_latency,
235 .maxlen = sizeof(unsigned int),
236 .mode = 0644,
237 .proc_handler = &proc_dointvec_minmax,
238 .strategy = &sysctl_intvec,
239 .extra1 = &min_sched_granularity_ns,
240 .extra2 = &max_sched_granularity_ns,
241 },
242 {
243 .ctl_name = CTL_UNNUMBERED,
244 .procname = "sched_wakeup_granularity_ns",
245 .data = &sysctl_sched_wakeup_granularity,
246 .maxlen = sizeof(unsigned int),
247 .mode = 0644,
248 .proc_handler = &proc_dointvec_minmax,
249 .strategy = &sysctl_intvec,
250 .extra1 = &min_wakeup_granularity_ns,
251 .extra2 = &max_wakeup_granularity_ns,
252 },
253 {
254 .ctl_name = CTL_UNNUMBERED,
255 .procname = "sched_batch_wakeup_granularity_ns",
256 .data = &sysctl_sched_batch_wakeup_granularity,
257 .maxlen = sizeof(unsigned int),
258 .mode = 0644,
259 .proc_handler = &proc_dointvec_minmax,
260 .strategy = &sysctl_intvec,
261 .extra1 = &min_wakeup_granularity_ns,
262 .extra2 = &max_wakeup_granularity_ns,
263 },
264 {
265 .ctl_name = CTL_UNNUMBERED,
266 .procname = "sched_child_runs_first",
267 .data = &sysctl_sched_child_runs_first,
268 .maxlen = sizeof(unsigned int),
269 .mode = 0644,
270 .proc_handler = &proc_dointvec,
271 },
272 {
273 .ctl_name = CTL_UNNUMBERED,
274 .procname = "sched_features",
275 .data = &sysctl_sched_features,
276 .maxlen = sizeof(unsigned int),
277 .mode = 0644,
278 .proc_handler = &proc_dointvec,
279 },
280 {
281 .ctl_name = CTL_UNNUMBERED,
282 .procname = "sched_migration_cost",
283 .data = &sysctl_sched_migration_cost,
284 .maxlen = sizeof(unsigned int),
285 .mode = 0644,
286 .proc_handler = &proc_dointvec,
287 },
288 #endif
289 {
290 .ctl_name = CTL_UNNUMBERED,
291 .procname = "sched_compat_yield",
292 .data = &sysctl_sched_compat_yield,
293 .maxlen = sizeof(unsigned int),
294 .mode = 0644,
295 .proc_handler = &proc_dointvec,
296 },
297 #ifdef CONFIG_PROVE_LOCKING
298 {
299 .ctl_name = CTL_UNNUMBERED,
300 .procname = "prove_locking",
301 .data = &prove_locking,
302 .maxlen = sizeof(int),
303 .mode = 0644,
304 .proc_handler = &proc_dointvec,
305 },
306 #endif
307 #ifdef CONFIG_LOCK_STAT
308 {
309 .ctl_name = CTL_UNNUMBERED,
310 .procname = "lock_stat",
311 .data = &lock_stat,
312 .maxlen = sizeof(int),
313 .mode = 0644,
314 .proc_handler = &proc_dointvec,
315 },
316 #endif
317 {
318 .ctl_name = KERN_PANIC,
319 .procname = "panic",
320 .data = &panic_timeout,
321 .maxlen = sizeof(int),
322 .mode = 0644,
323 .proc_handler = &proc_dointvec,
324 },
325 {
326 .ctl_name = KERN_CORE_USES_PID,
327 .procname = "core_uses_pid",
328 .data = &core_uses_pid,
329 .maxlen = sizeof(int),
330 .mode = 0644,
331 .proc_handler = &proc_dointvec,
332 },
333 #ifdef CONFIG_AUDITSYSCALL
334 {
335 .ctl_name = CTL_UNNUMBERED,
336 .procname = "audit_argv_kb",
337 .data = &audit_argv_kb,
338 .maxlen = sizeof(int),
339 .mode = 0644,
340 .proc_handler = &proc_dointvec,
341 },
342 #endif
343 {
344 .ctl_name = KERN_CORE_PATTERN,
345 .procname = "core_pattern",
346 .data = core_pattern,
347 .maxlen = CORENAME_MAX_SIZE,
348 .mode = 0644,
349 .proc_handler = &proc_dostring,
350 .strategy = &sysctl_string,
351 },
352 #ifdef CONFIG_PROC_SYSCTL
353 {
354 .ctl_name = KERN_TAINTED,
355 .procname = "tainted",
356 .data = &tainted,
357 .maxlen = sizeof(int),
358 .mode = 0644,
359 .proc_handler = &proc_dointvec_taint,
360 },
361 #endif
362 {
363 .ctl_name = KERN_CAP_BSET,
364 .procname = "cap-bound",
365 .data = &cap_bset,
366 .maxlen = sizeof(kernel_cap_t),
367 .mode = 0600,
368 .proc_handler = &proc_dointvec_bset,
369 },
370 #ifdef CONFIG_BLK_DEV_INITRD
371 {
372 .ctl_name = KERN_REALROOTDEV,
373 .procname = "real-root-dev",
374 .data = &real_root_dev,
375 .maxlen = sizeof(int),
376 .mode = 0644,
377 .proc_handler = &proc_dointvec,
378 },
379 #endif
380 {
381 .ctl_name = CTL_UNNUMBERED,
382 .procname = "print-fatal-signals",
383 .data = &print_fatal_signals,
384 .maxlen = sizeof(int),
385 .mode = 0644,
386 .proc_handler = &proc_dointvec,
387 },
388 #ifdef __sparc__
389 {
390 .ctl_name = KERN_SPARC_REBOOT,
391 .procname = "reboot-cmd",
392 .data = reboot_command,
393 .maxlen = 256,
394 .mode = 0644,
395 .proc_handler = &proc_dostring,
396 .strategy = &sysctl_string,
397 },
398 {
399 .ctl_name = KERN_SPARC_STOP_A,
400 .procname = "stop-a",
401 .data = &stop_a_enabled,
402 .maxlen = sizeof (int),
403 .mode = 0644,
404 .proc_handler = &proc_dointvec,
405 },
406 {
407 .ctl_name = KERN_SPARC_SCONS_PWROFF,
408 .procname = "scons-poweroff",
409 .data = &scons_pwroff,
410 .maxlen = sizeof (int),
411 .mode = 0644,
412 .proc_handler = &proc_dointvec,
413 },
414 #endif
415 #ifdef __hppa__
416 {
417 .ctl_name = KERN_HPPA_PWRSW,
418 .procname = "soft-power",
419 .data = &pwrsw_enabled,
420 .maxlen = sizeof (int),
421 .mode = 0644,
422 .proc_handler = &proc_dointvec,
423 },
424 {
425 .ctl_name = KERN_HPPA_UNALIGNED,
426 .procname = "unaligned-trap",
427 .data = &unaligned_enabled,
428 .maxlen = sizeof (int),
429 .mode = 0644,
430 .proc_handler = &proc_dointvec,
431 },
432 #endif
433 {
434 .ctl_name = KERN_CTLALTDEL,
435 .procname = "ctrl-alt-del",
436 .data = &C_A_D,
437 .maxlen = sizeof(int),
438 .mode = 0644,
439 .proc_handler = &proc_dointvec,
440 },
441 {
442 .ctl_name = KERN_PRINTK,
443 .procname = "printk",
444 .data = &console_loglevel,
445 .maxlen = 4*sizeof(int),
446 .mode = 0644,
447 .proc_handler = &proc_dointvec,
448 },
449 #ifdef CONFIG_KMOD
450 {
451 .ctl_name = KERN_MODPROBE,
452 .procname = "modprobe",
453 .data = &modprobe_path,
454 .maxlen = KMOD_PATH_LEN,
455 .mode = 0644,
456 .proc_handler = &proc_dostring,
457 .strategy = &sysctl_string,
458 },
459 #endif
460 #if defined(CONFIG_HOTPLUG) && defined(CONFIG_NET)
461 {
462 .ctl_name = KERN_HOTPLUG,
463 .procname = "hotplug",
464 .data = &uevent_helper,
465 .maxlen = UEVENT_HELPER_PATH_LEN,
466 .mode = 0644,
467 .proc_handler = &proc_dostring,
468 .strategy = &sysctl_string,
469 },
470 #endif
471 #ifdef CONFIG_CHR_DEV_SG
472 {
473 .ctl_name = KERN_SG_BIG_BUFF,
474 .procname = "sg-big-buff",
475 .data = &sg_big_buff,
476 .maxlen = sizeof (int),
477 .mode = 0444,
478 .proc_handler = &proc_dointvec,
479 },
480 #endif
481 #ifdef CONFIG_BSD_PROCESS_ACCT
482 {
483 .ctl_name = KERN_ACCT,
484 .procname = "acct",
485 .data = &acct_parm,
486 .maxlen = 3*sizeof(int),
487 .mode = 0644,
488 .proc_handler = &proc_dointvec,
489 },
490 #endif
491 #ifdef CONFIG_MAGIC_SYSRQ
492 {
493 .ctl_name = KERN_SYSRQ,
494 .procname = "sysrq",
495 .data = &__sysrq_enabled,
496 .maxlen = sizeof (int),
497 .mode = 0644,
498 .proc_handler = &proc_dointvec,
499 },
500 #endif
501 #ifdef CONFIG_PROC_SYSCTL
502 {
503 .ctl_name = KERN_CADPID,
504 .procname = "cad_pid",
505 .data = NULL,
506 .maxlen = sizeof (int),
507 .mode = 0600,
508 .proc_handler = &proc_do_cad_pid,
509 },
510 #endif
511 {
512 .ctl_name = KERN_MAX_THREADS,
513 .procname = "threads-max",
514 .data = &max_threads,
515 .maxlen = sizeof(int),
516 .mode = 0644,
517 .proc_handler = &proc_dointvec,
518 },
519 {
520 .ctl_name = KERN_RANDOM,
521 .procname = "random",
522 .mode = 0555,
523 .child = random_table,
524 },
525 #ifdef CONFIG_UNIX98_PTYS
526 {
527 .ctl_name = KERN_PTY,
528 .procname = "pty",
529 .mode = 0555,
530 .child = pty_table,
531 },
532 #endif
533 {
534 .ctl_name = KERN_OVERFLOWUID,
535 .procname = "overflowuid",
536 .data = &overflowuid,
537 .maxlen = sizeof(int),
538 .mode = 0644,
539 .proc_handler = &proc_dointvec_minmax,
540 .strategy = &sysctl_intvec,
541 .extra1 = &minolduid,
542 .extra2 = &maxolduid,
543 },
544 {
545 .ctl_name = KERN_OVERFLOWGID,
546 .procname = "overflowgid",
547 .data = &overflowgid,
548 .maxlen = sizeof(int),
549 .mode = 0644,
550 .proc_handler = &proc_dointvec_minmax,
551 .strategy = &sysctl_intvec,
552 .extra1 = &minolduid,
553 .extra2 = &maxolduid,
554 },
555 #ifdef CONFIG_S390
556 #ifdef CONFIG_MATHEMU
557 {
558 .ctl_name = KERN_IEEE_EMULATION_WARNINGS,
559 .procname = "ieee_emulation_warnings",
560 .data = &sysctl_ieee_emulation_warnings,
561 .maxlen = sizeof(int),
562 .mode = 0644,
563 .proc_handler = &proc_dointvec,
564 },
565 #endif
566 #ifdef CONFIG_NO_IDLE_HZ
567 {
568 .ctl_name = KERN_HZ_TIMER,
569 .procname = "hz_timer",
570 .data = &sysctl_hz_timer,
571 .maxlen = sizeof(int),
572 .mode = 0644,
573 .proc_handler = &proc_dointvec,
574 },
575 #endif
576 {
577 .ctl_name = KERN_S390_USER_DEBUG_LOGGING,
578 .procname = "userprocess_debug",
579 .data = &sysctl_userprocess_debug,
580 .maxlen = sizeof(int),
581 .mode = 0644,
582 .proc_handler = &proc_dointvec,
583 },
584 #endif
585 {
586 .ctl_name = KERN_PIDMAX,
587 .procname = "pid_max",
588 .data = &pid_max,
589 .maxlen = sizeof (int),
590 .mode = 0644,
591 .proc_handler = &proc_dointvec_minmax,
592 .strategy = sysctl_intvec,
593 .extra1 = &pid_max_min,
594 .extra2 = &pid_max_max,
595 },
596 {
597 .ctl_name = KERN_PANIC_ON_OOPS,
598 .procname = "panic_on_oops",
599 .data = &panic_on_oops,
600 .maxlen = sizeof(int),
601 .mode = 0644,
602 .proc_handler = &proc_dointvec,
603 },
604 {
605 .ctl_name = KERN_PRINTK_RATELIMIT,
606 .procname = "printk_ratelimit",
607 .data = &printk_ratelimit_jiffies,
608 .maxlen = sizeof(int),
609 .mode = 0644,
610 .proc_handler = &proc_dointvec_jiffies,
611 .strategy = &sysctl_jiffies,
612 },
613 {
614 .ctl_name = KERN_PRINTK_RATELIMIT_BURST,
615 .procname = "printk_ratelimit_burst",
616 .data = &printk_ratelimit_burst,
617 .maxlen = sizeof(int),
618 .mode = 0644,
619 .proc_handler = &proc_dointvec,
620 },
621 {
622 .ctl_name = KERN_NGROUPS_MAX,
623 .procname = "ngroups_max",
624 .data = &ngroups_max,
625 .maxlen = sizeof (int),
626 .mode = 0444,
627 .proc_handler = &proc_dointvec,
628 },
629 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_X86)
630 {
631 .ctl_name = KERN_UNKNOWN_NMI_PANIC,
632 .procname = "unknown_nmi_panic",
633 .data = &unknown_nmi_panic,
634 .maxlen = sizeof (int),
635 .mode = 0644,
636 .proc_handler = &proc_dointvec,
637 },
638 {
639 .ctl_name = KERN_NMI_WATCHDOG,
640 .procname = "nmi_watchdog",
641 .data = &nmi_watchdog_enabled,
642 .maxlen = sizeof (int),
643 .mode = 0644,
644 .proc_handler = &proc_nmi_enabled,
645 },
646 #endif
647 #if defined(CONFIG_X86)
648 {
649 .ctl_name = KERN_PANIC_ON_NMI,
650 .procname = "panic_on_unrecovered_nmi",
651 .data = &panic_on_unrecovered_nmi,
652 .maxlen = sizeof(int),
653 .mode = 0644,
654 .proc_handler = &proc_dointvec,
655 },
656 {
657 .ctl_name = KERN_BOOTLOADER_TYPE,
658 .procname = "bootloader_type",
659 .data = &bootloader_type,
660 .maxlen = sizeof (int),
661 .mode = 0444,
662 .proc_handler = &proc_dointvec,
663 },
664 {
665 .ctl_name = CTL_UNNUMBERED,
666 .procname = "kstack_depth_to_print",
667 .data = &kstack_depth_to_print,
668 .maxlen = sizeof(int),
669 .mode = 0644,
670 .proc_handler = &proc_dointvec,
671 },
672 #endif
673 #if defined(CONFIG_MMU)
674 {
675 .ctl_name = KERN_RANDOMIZE,
676 .procname = "randomize_va_space",
677 .data = &randomize_va_space,
678 .maxlen = sizeof(int),
679 .mode = 0644,
680 .proc_handler = &proc_dointvec,
681 },
682 #endif
683 #if defined(CONFIG_S390) && defined(CONFIG_SMP)
684 {
685 .ctl_name = KERN_SPIN_RETRY,
686 .procname = "spin_retry",
687 .data = &spin_retry,
688 .maxlen = sizeof (int),
689 .mode = 0644,
690 .proc_handler = &proc_dointvec,
691 },
692 #endif
693 #if defined(CONFIG_ACPI_SLEEP) && defined(CONFIG_X86)
694 {
695 .ctl_name = KERN_ACPI_VIDEO_FLAGS,
696 .procname = "acpi_video_flags",
697 .data = &acpi_realmode_flags,
698 .maxlen = sizeof (unsigned long),
699 .mode = 0644,
700 .proc_handler = &proc_doulongvec_minmax,
701 },
702 #endif
703 #ifdef CONFIG_IA64
704 {
705 .ctl_name = KERN_IA64_UNALIGNED,
706 .procname = "ignore-unaligned-usertrap",
707 .data = &no_unaligned_warning,
708 .maxlen = sizeof (int),
709 .mode = 0644,
710 .proc_handler = &proc_dointvec,
711 },
712 #endif
713 #ifdef CONFIG_COMPAT
714 {
715 .ctl_name = KERN_COMPAT_LOG,
716 .procname = "compat-log",
717 .data = &compat_log,
718 .maxlen = sizeof (int),
719 .mode = 0644,
720 .proc_handler = &proc_dointvec,
721 },
722 #endif
723 #ifdef CONFIG_RT_MUTEXES
724 {
725 .ctl_name = KERN_MAX_LOCK_DEPTH,
726 .procname = "max_lock_depth",
727 .data = &max_lock_depth,
728 .maxlen = sizeof(int),
729 .mode = 0644,
730 .proc_handler = &proc_dointvec,
731 },
732 #endif
733 #ifdef CONFIG_PROC_FS
734 {
735 .ctl_name = CTL_UNNUMBERED,
736 .procname = "maps_protect",
737 .data = &maps_protect,
738 .maxlen = sizeof(int),
739 .mode = 0644,
740 .proc_handler = &proc_dointvec,
741 },
742 #endif
743 {
744 .ctl_name = CTL_UNNUMBERED,
745 .procname = "poweroff_cmd",
746 .data = &poweroff_cmd,
747 .maxlen = POWEROFF_CMD_PATH_LEN,
748 .mode = 0644,
749 .proc_handler = &proc_dostring,
750 .strategy = &sysctl_string,
751 },
752 /*
753 * NOTE: do not add new entries to this table unless you have read
754 * Documentation/sysctl/ctl_unnumbered.txt
755 */
756 { .ctl_name = 0 }
757 };
758
759 /* Constants for minimum and maximum testing in vm_table.
760 We use these as one-element integer vectors. */
761 static int zero;
762 static int two = 2;
763 static int one_hundred = 100;
764
765
766 static ctl_table vm_table[] = {
767 {
768 .ctl_name = VM_OVERCOMMIT_MEMORY,
769 .procname = "overcommit_memory",
770 .data = &sysctl_overcommit_memory,
771 .maxlen = sizeof(sysctl_overcommit_memory),
772 .mode = 0644,
773 .proc_handler = &proc_dointvec,
774 },
775 {
776 .ctl_name = VM_PANIC_ON_OOM,
777 .procname = "panic_on_oom",
778 .data = &sysctl_panic_on_oom,
779 .maxlen = sizeof(sysctl_panic_on_oom),
780 .mode = 0644,
781 .proc_handler = &proc_dointvec,
782 },
783 {
784 .ctl_name = VM_OVERCOMMIT_RATIO,
785 .procname = "overcommit_ratio",
786 .data = &sysctl_overcommit_ratio,
787 .maxlen = sizeof(sysctl_overcommit_ratio),
788 .mode = 0644,
789 .proc_handler = &proc_dointvec,
790 },
791 {
792 .ctl_name = VM_PAGE_CLUSTER,
793 .procname = "page-cluster",
794 .data = &page_cluster,
795 .maxlen = sizeof(int),
796 .mode = 0644,
797 .proc_handler = &proc_dointvec,
798 },
799 {
800 .ctl_name = VM_DIRTY_BACKGROUND,
801 .procname = "dirty_background_ratio",
802 .data = &dirty_background_ratio,
803 .maxlen = sizeof(dirty_background_ratio),
804 .mode = 0644,
805 .proc_handler = &proc_dointvec_minmax,
806 .strategy = &sysctl_intvec,
807 .extra1 = &zero,
808 .extra2 = &one_hundred,
809 },
810 {
811 .ctl_name = VM_DIRTY_RATIO,
812 .procname = "dirty_ratio",
813 .data = &vm_dirty_ratio,
814 .maxlen = sizeof(vm_dirty_ratio),
815 .mode = 0644,
816 .proc_handler = &dirty_ratio_handler,
817 .strategy = &sysctl_intvec,
818 .extra1 = &zero,
819 .extra2 = &one_hundred,
820 },
821 {
822 .ctl_name = VM_DIRTY_WB_CS,
823 .procname = "dirty_writeback_centisecs",
824 .data = &dirty_writeback_interval,
825 .maxlen = sizeof(dirty_writeback_interval),
826 .mode = 0644,
827 .proc_handler = &dirty_writeback_centisecs_handler,
828 },
829 {
830 .ctl_name = VM_DIRTY_EXPIRE_CS,
831 .procname = "dirty_expire_centisecs",
832 .data = &dirty_expire_interval,
833 .maxlen = sizeof(dirty_expire_interval),
834 .mode = 0644,
835 .proc_handler = &proc_dointvec_userhz_jiffies,
836 },
837 {
838 .ctl_name = VM_NR_PDFLUSH_THREADS,
839 .procname = "nr_pdflush_threads",
840 .data = &nr_pdflush_threads,
841 .maxlen = sizeof nr_pdflush_threads,
842 .mode = 0444 /* read-only*/,
843 .proc_handler = &proc_dointvec,
844 },
845 {
846 .ctl_name = VM_SWAPPINESS,
847 .procname = "swappiness",
848 .data = &vm_swappiness,
849 .maxlen = sizeof(vm_swappiness),
850 .mode = 0644,
851 .proc_handler = &proc_dointvec_minmax,
852 .strategy = &sysctl_intvec,
853 .extra1 = &zero,
854 .extra2 = &one_hundred,
855 },
856 #ifdef CONFIG_HUGETLB_PAGE
857 {
858 .ctl_name = VM_HUGETLB_PAGES,
859 .procname = "nr_hugepages",
860 .data = &max_huge_pages,
861 .maxlen = sizeof(unsigned long),
862 .mode = 0644,
863 .proc_handler = &hugetlb_sysctl_handler,
864 .extra1 = (void *)&hugetlb_zero,
865 .extra2 = (void *)&hugetlb_infinity,
866 },
867 {
868 .ctl_name = VM_HUGETLB_GROUP,
869 .procname = "hugetlb_shm_group",
870 .data = &sysctl_hugetlb_shm_group,
871 .maxlen = sizeof(gid_t),
872 .mode = 0644,
873 .proc_handler = &proc_dointvec,
874 },
875 {
876 .ctl_name = CTL_UNNUMBERED,
877 .procname = "hugepages_treat_as_movable",
878 .data = &hugepages_treat_as_movable,
879 .maxlen = sizeof(int),
880 .mode = 0644,
881 .proc_handler = &hugetlb_treat_movable_handler,
882 },
883 {
884 .ctl_name = CTL_UNNUMBERED,
885 .procname = "hugetlb_dynamic_pool",
886 .data = &hugetlb_dynamic_pool,
887 .maxlen = sizeof(hugetlb_dynamic_pool),
888 .mode = 0644,
889 .proc_handler = &proc_dointvec,
890 },
891 #endif
892 {
893 .ctl_name = VM_LOWMEM_RESERVE_RATIO,
894 .procname = "lowmem_reserve_ratio",
895 .data = &sysctl_lowmem_reserve_ratio,
896 .maxlen = sizeof(sysctl_lowmem_reserve_ratio),
897 .mode = 0644,
898 .proc_handler = &lowmem_reserve_ratio_sysctl_handler,
899 .strategy = &sysctl_intvec,
900 },
901 {
902 .ctl_name = VM_DROP_PAGECACHE,
903 .procname = "drop_caches",
904 .data = &sysctl_drop_caches,
905 .maxlen = sizeof(int),
906 .mode = 0644,
907 .proc_handler = drop_caches_sysctl_handler,
908 .strategy = &sysctl_intvec,
909 },
910 {
911 .ctl_name = VM_MIN_FREE_KBYTES,
912 .procname = "min_free_kbytes",
913 .data = &min_free_kbytes,
914 .maxlen = sizeof(min_free_kbytes),
915 .mode = 0644,
916 .proc_handler = &min_free_kbytes_sysctl_handler,
917 .strategy = &sysctl_intvec,
918 .extra1 = &zero,
919 },
920 {
921 .ctl_name = VM_PERCPU_PAGELIST_FRACTION,
922 .procname = "percpu_pagelist_fraction",
923 .data = &percpu_pagelist_fraction,
924 .maxlen = sizeof(percpu_pagelist_fraction),
925 .mode = 0644,
926 .proc_handler = &percpu_pagelist_fraction_sysctl_handler,
927 .strategy = &sysctl_intvec,
928 .extra1 = &min_percpu_pagelist_fract,
929 },
930 #ifdef CONFIG_MMU
931 {
932 .ctl_name = VM_MAX_MAP_COUNT,
933 .procname = "max_map_count",
934 .data = &sysctl_max_map_count,
935 .maxlen = sizeof(sysctl_max_map_count),
936 .mode = 0644,
937 .proc_handler = &proc_dointvec
938 },
939 #endif
940 {
941 .ctl_name = VM_LAPTOP_MODE,
942 .procname = "laptop_mode",
943 .data = &laptop_mode,
944 .maxlen = sizeof(laptop_mode),
945 .mode = 0644,
946 .proc_handler = &proc_dointvec_jiffies,
947 .strategy = &sysctl_jiffies,
948 },
949 {
950 .ctl_name = VM_BLOCK_DUMP,
951 .procname = "block_dump",
952 .data = &block_dump,
953 .maxlen = sizeof(block_dump),
954 .mode = 0644,
955 .proc_handler = &proc_dointvec,
956 .strategy = &sysctl_intvec,
957 .extra1 = &zero,
958 },
959 {
960 .ctl_name = VM_VFS_CACHE_PRESSURE,
961 .procname = "vfs_cache_pressure",
962 .data = &sysctl_vfs_cache_pressure,
963 .maxlen = sizeof(sysctl_vfs_cache_pressure),
964 .mode = 0644,
965 .proc_handler = &proc_dointvec,
966 .strategy = &sysctl_intvec,
967 .extra1 = &zero,
968 },
969 #ifdef HAVE_ARCH_PICK_MMAP_LAYOUT
970 {
971 .ctl_name = VM_LEGACY_VA_LAYOUT,
972 .procname = "legacy_va_layout",
973 .data = &sysctl_legacy_va_layout,
974 .maxlen = sizeof(sysctl_legacy_va_layout),
975 .mode = 0644,
976 .proc_handler = &proc_dointvec,
977 .strategy = &sysctl_intvec,
978 .extra1 = &zero,
979 },
980 #endif
981 #ifdef CONFIG_NUMA
982 {
983 .ctl_name = VM_ZONE_RECLAIM_MODE,
984 .procname = "zone_reclaim_mode",
985 .data = &zone_reclaim_mode,
986 .maxlen = sizeof(zone_reclaim_mode),
987 .mode = 0644,
988 .proc_handler = &proc_dointvec,
989 .strategy = &sysctl_intvec,
990 .extra1 = &zero,
991 },
992 {
993 .ctl_name = VM_MIN_UNMAPPED,
994 .procname = "min_unmapped_ratio",
995 .data = &sysctl_min_unmapped_ratio,
996 .maxlen = sizeof(sysctl_min_unmapped_ratio),
997 .mode = 0644,
998 .proc_handler = &sysctl_min_unmapped_ratio_sysctl_handler,
999 .strategy = &sysctl_intvec,
1000 .extra1 = &zero,
1001 .extra2 = &one_hundred,
1002 },
1003 {
1004 .ctl_name = VM_MIN_SLAB,
1005 .procname = "min_slab_ratio",
1006 .data = &sysctl_min_slab_ratio,
1007 .maxlen = sizeof(sysctl_min_slab_ratio),
1008 .mode = 0644,
1009 .proc_handler = &sysctl_min_slab_ratio_sysctl_handler,
1010 .strategy = &sysctl_intvec,
1011 .extra1 = &zero,
1012 .extra2 = &one_hundred,
1013 },
1014 #endif
1015 #ifdef CONFIG_SMP
1016 {
1017 .ctl_name = CTL_UNNUMBERED,
1018 .procname = "stat_interval",
1019 .data = &sysctl_stat_interval,
1020 .maxlen = sizeof(sysctl_stat_interval),
1021 .mode = 0644,
1022 .proc_handler = &proc_dointvec_jiffies,
1023 .strategy = &sysctl_jiffies,
1024 },
1025 #endif
1026 #ifdef CONFIG_SECURITY
1027 {
1028 .ctl_name = CTL_UNNUMBERED,
1029 .procname = "mmap_min_addr",
1030 .data = &mmap_min_addr,
1031 .maxlen = sizeof(unsigned long),
1032 .mode = 0644,
1033 .proc_handler = &proc_doulongvec_minmax,
1034 },
1035 #endif
1036 #ifdef CONFIG_NUMA
1037 {
1038 .ctl_name = CTL_UNNUMBERED,
1039 .procname = "numa_zonelist_order",
1040 .data = &numa_zonelist_order,
1041 .maxlen = NUMA_ZONELIST_ORDER_LEN,
1042 .mode = 0644,
1043 .proc_handler = &numa_zonelist_order_handler,
1044 .strategy = &sysctl_string,
1045 },
1046 #endif
1047 #if (defined(CONFIG_X86_32) && !defined(CONFIG_UML))|| \
1048 (defined(CONFIG_SUPERH) && defined(CONFIG_VSYSCALL))
1049 {
1050 .ctl_name = VM_VDSO_ENABLED,
1051 .procname = "vdso_enabled",
1052 .data = &vdso_enabled,
1053 .maxlen = sizeof(vdso_enabled),
1054 .mode = 0644,
1055 .proc_handler = &proc_dointvec,
1056 .strategy = &sysctl_intvec,
1057 .extra1 = &zero,
1058 },
1059 #endif
1060 /*
1061 * NOTE: do not add new entries to this table unless you have read
1062 * Documentation/sysctl/ctl_unnumbered.txt
1063 */
1064 { .ctl_name = 0 }
1065 };
1066
1067 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1068 static ctl_table binfmt_misc_table[] = {
1069 { .ctl_name = 0 }
1070 };
1071 #endif
1072
1073 static ctl_table fs_table[] = {
1074 {
1075 .ctl_name = FS_NRINODE,
1076 .procname = "inode-nr",
1077 .data = &inodes_stat,
1078 .maxlen = 2*sizeof(int),
1079 .mode = 0444,
1080 .proc_handler = &proc_dointvec,
1081 },
1082 {
1083 .ctl_name = FS_STATINODE,
1084 .procname = "inode-state",
1085 .data = &inodes_stat,
1086 .maxlen = 7*sizeof(int),
1087 .mode = 0444,
1088 .proc_handler = &proc_dointvec,
1089 },
1090 {
1091 .ctl_name = FS_NRFILE,
1092 .procname = "file-nr",
1093 .data = &files_stat,
1094 .maxlen = 3*sizeof(int),
1095 .mode = 0444,
1096 .proc_handler = &proc_nr_files,
1097 },
1098 {
1099 .ctl_name = FS_MAXFILE,
1100 .procname = "file-max",
1101 .data = &files_stat.max_files,
1102 .maxlen = sizeof(int),
1103 .mode = 0644,
1104 .proc_handler = &proc_dointvec,
1105 },
1106 {
1107 .ctl_name = FS_DENTRY,
1108 .procname = "dentry-state",
1109 .data = &dentry_stat,
1110 .maxlen = 6*sizeof(int),
1111 .mode = 0444,
1112 .proc_handler = &proc_dointvec,
1113 },
1114 {
1115 .ctl_name = FS_OVERFLOWUID,
1116 .procname = "overflowuid",
1117 .data = &fs_overflowuid,
1118 .maxlen = sizeof(int),
1119 .mode = 0644,
1120 .proc_handler = &proc_dointvec_minmax,
1121 .strategy = &sysctl_intvec,
1122 .extra1 = &minolduid,
1123 .extra2 = &maxolduid,
1124 },
1125 {
1126 .ctl_name = FS_OVERFLOWGID,
1127 .procname = "overflowgid",
1128 .data = &fs_overflowgid,
1129 .maxlen = sizeof(int),
1130 .mode = 0644,
1131 .proc_handler = &proc_dointvec_minmax,
1132 .strategy = &sysctl_intvec,
1133 .extra1 = &minolduid,
1134 .extra2 = &maxolduid,
1135 },
1136 {
1137 .ctl_name = FS_LEASES,
1138 .procname = "leases-enable",
1139 .data = &leases_enable,
1140 .maxlen = sizeof(int),
1141 .mode = 0644,
1142 .proc_handler = &proc_dointvec,
1143 },
1144 #ifdef CONFIG_DNOTIFY
1145 {
1146 .ctl_name = FS_DIR_NOTIFY,
1147 .procname = "dir-notify-enable",
1148 .data = &dir_notify_enable,
1149 .maxlen = sizeof(int),
1150 .mode = 0644,
1151 .proc_handler = &proc_dointvec,
1152 },
1153 #endif
1154 #ifdef CONFIG_MMU
1155 {
1156 .ctl_name = FS_LEASE_TIME,
1157 .procname = "lease-break-time",
1158 .data = &lease_break_time,
1159 .maxlen = sizeof(int),
1160 .mode = 0644,
1161 .proc_handler = &proc_dointvec_minmax,
1162 .strategy = &sysctl_intvec,
1163 .extra1 = &zero,
1164 .extra2 = &two,
1165 },
1166 {
1167 .ctl_name = FS_AIO_NR,
1168 .procname = "aio-nr",
1169 .data = &aio_nr,
1170 .maxlen = sizeof(aio_nr),
1171 .mode = 0444,
1172 .proc_handler = &proc_doulongvec_minmax,
1173 },
1174 {
1175 .ctl_name = FS_AIO_MAX_NR,
1176 .procname = "aio-max-nr",
1177 .data = &aio_max_nr,
1178 .maxlen = sizeof(aio_max_nr),
1179 .mode = 0644,
1180 .proc_handler = &proc_doulongvec_minmax,
1181 },
1182 #ifdef CONFIG_INOTIFY_USER
1183 {
1184 .ctl_name = FS_INOTIFY,
1185 .procname = "inotify",
1186 .mode = 0555,
1187 .child = inotify_table,
1188 },
1189 #endif
1190 #endif
1191 {
1192 .ctl_name = KERN_SETUID_DUMPABLE,
1193 .procname = "suid_dumpable",
1194 .data = &suid_dumpable,
1195 .maxlen = sizeof(int),
1196 .mode = 0644,
1197 .proc_handler = &proc_dointvec,
1198 },
1199 #if defined(CONFIG_BINFMT_MISC) || defined(CONFIG_BINFMT_MISC_MODULE)
1200 {
1201 .ctl_name = CTL_UNNUMBERED,
1202 .procname = "binfmt_misc",
1203 .mode = 0555,
1204 .child = binfmt_misc_table,
1205 },
1206 #endif
1207 /*
1208 * NOTE: do not add new entries to this table unless you have read
1209 * Documentation/sysctl/ctl_unnumbered.txt
1210 */
1211 { .ctl_name = 0 }
1212 };
1213
1214 static ctl_table debug_table[] = {
1215 #if defined(CONFIG_X86) || defined(CONFIG_PPC)
1216 {
1217 .ctl_name = CTL_UNNUMBERED,
1218 .procname = "exception-trace",
1219 .data = &show_unhandled_signals,
1220 .maxlen = sizeof(int),
1221 .mode = 0644,
1222 .proc_handler = proc_dointvec
1223 },
1224 #endif
1225 { .ctl_name = 0 }
1226 };
1227
1228 static ctl_table dev_table[] = {
1229 { .ctl_name = 0 }
1230 };
1231
1232 static DEFINE_SPINLOCK(sysctl_lock);
1233
1234 /* called under sysctl_lock */
1235 static int use_table(struct ctl_table_header *p)
1236 {
1237 if (unlikely(p->unregistering))
1238 return 0;
1239 p->used++;
1240 return 1;
1241 }
1242
1243 /* called under sysctl_lock */
1244 static void unuse_table(struct ctl_table_header *p)
1245 {
1246 if (!--p->used)
1247 if (unlikely(p->unregistering))
1248 complete(p->unregistering);
1249 }
1250
1251 /* called under sysctl_lock, will reacquire if has to wait */
1252 static void start_unregistering(struct ctl_table_header *p)
1253 {
1254 /*
1255 * if p->used is 0, nobody will ever touch that entry again;
1256 * we'll eliminate all paths to it before dropping sysctl_lock
1257 */
1258 if (unlikely(p->used)) {
1259 struct completion wait;
1260 init_completion(&wait);
1261 p->unregistering = &wait;
1262 spin_unlock(&sysctl_lock);
1263 wait_for_completion(&wait);
1264 spin_lock(&sysctl_lock);
1265 }
1266 /*
1267 * do not remove from the list until nobody holds it; walking the
1268 * list in do_sysctl() relies on that.
1269 */
1270 list_del_init(&p->ctl_entry);
1271 }
1272
1273 void sysctl_head_finish(struct ctl_table_header *head)
1274 {
1275 if (!head)
1276 return;
1277 spin_lock(&sysctl_lock);
1278 unuse_table(head);
1279 spin_unlock(&sysctl_lock);
1280 }
1281
1282 struct ctl_table_header *sysctl_head_next(struct ctl_table_header *prev)
1283 {
1284 struct ctl_table_header *head;
1285 struct list_head *tmp;
1286 spin_lock(&sysctl_lock);
1287 if (prev) {
1288 tmp = &prev->ctl_entry;
1289 unuse_table(prev);
1290 goto next;
1291 }
1292 tmp = &root_table_header.ctl_entry;
1293 for (;;) {
1294 head = list_entry(tmp, struct ctl_table_header, ctl_entry);
1295
1296 if (!use_table(head))
1297 goto next;
1298 spin_unlock(&sysctl_lock);
1299 return head;
1300 next:
1301 tmp = tmp->next;
1302 if (tmp == &root_table_header.ctl_entry)
1303 break;
1304 }
1305 spin_unlock(&sysctl_lock);
1306 return NULL;
1307 }
1308
1309 #ifdef CONFIG_SYSCTL_SYSCALL
1310 int do_sysctl(int __user *name, int nlen, void __user *oldval, size_t __user *oldlenp,
1311 void __user *newval, size_t newlen)
1312 {
1313 struct ctl_table_header *head;
1314 int error = -ENOTDIR;
1315
1316 if (nlen <= 0 || nlen >= CTL_MAXNAME)
1317 return -ENOTDIR;
1318 if (oldval) {
1319 int old_len;
1320 if (!oldlenp || get_user(old_len, oldlenp))
1321 return -EFAULT;
1322 }
1323
1324 for (head = sysctl_head_next(NULL); head;
1325 head = sysctl_head_next(head)) {
1326 error = parse_table(name, nlen, oldval, oldlenp,
1327 newval, newlen, head->ctl_table);
1328 if (error != -ENOTDIR) {
1329 sysctl_head_finish(head);
1330 break;
1331 }
1332 }
1333 return error;
1334 }
1335
1336 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
1337 {
1338 struct __sysctl_args tmp;
1339 int error;
1340
1341 if (copy_from_user(&tmp, args, sizeof(tmp)))
1342 return -EFAULT;
1343
1344 lock_kernel();
1345 error = do_sysctl(tmp.name, tmp.nlen, tmp.oldval, tmp.oldlenp,
1346 tmp.newval, tmp.newlen);
1347 unlock_kernel();
1348 return error;
1349 }
1350 #endif /* CONFIG_SYSCTL_SYSCALL */
1351
1352 /*
1353 * sysctl_perm does NOT grant the superuser all rights automatically, because
1354 * some sysctl variables are readonly even to root.
1355 */
1356
1357 static int test_perm(int mode, int op)
1358 {
1359 if (!current->euid)
1360 mode >>= 6;
1361 else if (in_egroup_p(0))
1362 mode >>= 3;
1363 if ((mode & op & 0007) == op)
1364 return 0;
1365 return -EACCES;
1366 }
1367
1368 int sysctl_perm(ctl_table *table, int op)
1369 {
1370 int error;
1371 error = security_sysctl(table, op);
1372 if (error)
1373 return error;
1374 return test_perm(table->mode, op);
1375 }
1376
1377 #ifdef CONFIG_SYSCTL_SYSCALL
1378 static int parse_table(int __user *name, int nlen,
1379 void __user *oldval, size_t __user *oldlenp,
1380 void __user *newval, size_t newlen,
1381 ctl_table *table)
1382 {
1383 int n;
1384 repeat:
1385 if (!nlen)
1386 return -ENOTDIR;
1387 if (get_user(n, name))
1388 return -EFAULT;
1389 for ( ; table->ctl_name || table->procname; table++) {
1390 if (!table->ctl_name)
1391 continue;
1392 if (n == table->ctl_name) {
1393 int error;
1394 if (table->child) {
1395 if (sysctl_perm(table, 001))
1396 return -EPERM;
1397 name++;
1398 nlen--;
1399 table = table->child;
1400 goto repeat;
1401 }
1402 error = do_sysctl_strategy(table, name, nlen,
1403 oldval, oldlenp,
1404 newval, newlen);
1405 return error;
1406 }
1407 }
1408 return -ENOTDIR;
1409 }
1410
1411 /* Perform the actual read/write of a sysctl table entry. */
1412 int do_sysctl_strategy (ctl_table *table,
1413 int __user *name, int nlen,
1414 void __user *oldval, size_t __user *oldlenp,
1415 void __user *newval, size_t newlen)
1416 {
1417 int op = 0, rc;
1418 size_t len;
1419
1420 if (oldval)
1421 op |= 004;
1422 if (newval)
1423 op |= 002;
1424 if (sysctl_perm(table, op))
1425 return -EPERM;
1426
1427 if (table->strategy) {
1428 rc = table->strategy(table, name, nlen, oldval, oldlenp,
1429 newval, newlen);
1430 if (rc < 0)
1431 return rc;
1432 if (rc > 0)
1433 return 0;
1434 }
1435
1436 /* If there is no strategy routine, or if the strategy returns
1437 * zero, proceed with automatic r/w */
1438 if (table->data && table->maxlen) {
1439 if (oldval && oldlenp) {
1440 if (get_user(len, oldlenp))
1441 return -EFAULT;
1442 if (len) {
1443 if (len > table->maxlen)
1444 len = table->maxlen;
1445 if(copy_to_user(oldval, table->data, len))
1446 return -EFAULT;
1447 if(put_user(len, oldlenp))
1448 return -EFAULT;
1449 }
1450 }
1451 if (newval && newlen) {
1452 len = newlen;
1453 if (len > table->maxlen)
1454 len = table->maxlen;
1455 if(copy_from_user(table->data, newval, len))
1456 return -EFAULT;
1457 }
1458 }
1459 return 0;
1460 }
1461 #endif /* CONFIG_SYSCTL_SYSCALL */
1462
1463 static void sysctl_set_parent(struct ctl_table *parent, struct ctl_table *table)
1464 {
1465 for (; table->ctl_name || table->procname; table++) {
1466 table->parent = parent;
1467 if (table->child)
1468 sysctl_set_parent(table, table->child);
1469 }
1470 }
1471
1472 static __init int sysctl_init(void)
1473 {
1474 sysctl_set_parent(NULL, root_table);
1475 return 0;
1476 }
1477
1478 core_initcall(sysctl_init);
1479
1480 /**
1481 * register_sysctl_table - register a sysctl hierarchy
1482 * @table: the top-level table structure
1483 *
1484 * Register a sysctl table hierarchy. @table should be a filled in ctl_table
1485 * array. An entry with a ctl_name of 0 terminates the table.
1486 *
1487 * The members of the &ctl_table structure are used as follows:
1488 *
1489 * ctl_name - This is the numeric sysctl value used by sysctl(2). The number
1490 * must be unique within that level of sysctl
1491 *
1492 * procname - the name of the sysctl file under /proc/sys. Set to %NULL to not
1493 * enter a sysctl file
1494 *
1495 * data - a pointer to data for use by proc_handler
1496 *
1497 * maxlen - the maximum size in bytes of the data
1498 *
1499 * mode - the file permissions for the /proc/sys file, and for sysctl(2)
1500 *
1501 * child - a pointer to the child sysctl table if this entry is a directory, or
1502 * %NULL.
1503 *
1504 * proc_handler - the text handler routine (described below)
1505 *
1506 * strategy - the strategy routine (described below)
1507 *
1508 * de - for internal use by the sysctl routines
1509 *
1510 * extra1, extra2 - extra pointers usable by the proc handler routines
1511 *
1512 * Leaf nodes in the sysctl tree will be represented by a single file
1513 * under /proc; non-leaf nodes will be represented by directories.
1514 *
1515 * sysctl(2) can automatically manage read and write requests through
1516 * the sysctl table. The data and maxlen fields of the ctl_table
1517 * struct enable minimal validation of the values being written to be
1518 * performed, and the mode field allows minimal authentication.
1519 *
1520 * More sophisticated management can be enabled by the provision of a
1521 * strategy routine with the table entry. This will be called before
1522 * any automatic read or write of the data is performed.
1523 *
1524 * The strategy routine may return
1525 *
1526 * < 0 - Error occurred (error is passed to user process)
1527 *
1528 * 0 - OK - proceed with automatic read or write.
1529 *
1530 * > 0 - OK - read or write has been done by the strategy routine, so
1531 * return immediately.
1532 *
1533 * There must be a proc_handler routine for any terminal nodes
1534 * mirrored under /proc/sys (non-terminals are handled by a built-in
1535 * directory handler). Several default handlers are available to
1536 * cover common cases -
1537 *
1538 * proc_dostring(), proc_dointvec(), proc_dointvec_jiffies(),
1539 * proc_dointvec_userhz_jiffies(), proc_dointvec_minmax(),
1540 * proc_doulongvec_ms_jiffies_minmax(), proc_doulongvec_minmax()
1541 *
1542 * It is the handler's job to read the input buffer from user memory
1543 * and process it. The handler should return 0 on success.
1544 *
1545 * This routine returns %NULL on a failure to register, and a pointer
1546 * to the table header on success.
1547 */
1548 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1549 {
1550 struct ctl_table_header *tmp;
1551 tmp = kmalloc(sizeof(struct ctl_table_header), GFP_KERNEL);
1552 if (!tmp)
1553 return NULL;
1554 tmp->ctl_table = table;
1555 INIT_LIST_HEAD(&tmp->ctl_entry);
1556 tmp->used = 0;
1557 tmp->unregistering = NULL;
1558 sysctl_set_parent(NULL, table);
1559 spin_lock(&sysctl_lock);
1560 list_add_tail(&tmp->ctl_entry, &root_table_header.ctl_entry);
1561 spin_unlock(&sysctl_lock);
1562 return tmp;
1563 }
1564
1565 /**
1566 * unregister_sysctl_table - unregister a sysctl table hierarchy
1567 * @header: the header returned from register_sysctl_table
1568 *
1569 * Unregisters the sysctl table and all children. proc entries may not
1570 * actually be removed until they are no longer used by anyone.
1571 */
1572 void unregister_sysctl_table(struct ctl_table_header * header)
1573 {
1574 might_sleep();
1575 spin_lock(&sysctl_lock);
1576 start_unregistering(header);
1577 spin_unlock(&sysctl_lock);
1578 kfree(header);
1579 }
1580
1581 #else /* !CONFIG_SYSCTL */
1582 struct ctl_table_header *register_sysctl_table(ctl_table * table)
1583 {
1584 return NULL;
1585 }
1586
1587 void unregister_sysctl_table(struct ctl_table_header * table)
1588 {
1589 }
1590
1591 #endif /* CONFIG_SYSCTL */
1592
1593 /*
1594 * /proc/sys support
1595 */
1596
1597 #ifdef CONFIG_PROC_SYSCTL
1598
1599 static int _proc_do_string(void* data, int maxlen, int write,
1600 struct file *filp, void __user *buffer,
1601 size_t *lenp, loff_t *ppos)
1602 {
1603 size_t len;
1604 char __user *p;
1605 char c;
1606
1607 if (!data || !maxlen || !*lenp) {
1608 *lenp = 0;
1609 return 0;
1610 }
1611
1612 if (write) {
1613 len = 0;
1614 p = buffer;
1615 while (len < *lenp) {
1616 if (get_user(c, p++))
1617 return -EFAULT;
1618 if (c == 0 || c == '\n')
1619 break;
1620 len++;
1621 }
1622 if (len >= maxlen)
1623 len = maxlen-1;
1624 if(copy_from_user(data, buffer, len))
1625 return -EFAULT;
1626 ((char *) data)[len] = 0;
1627 *ppos += *lenp;
1628 } else {
1629 len = strlen(data);
1630 if (len > maxlen)
1631 len = maxlen;
1632
1633 if (*ppos > len) {
1634 *lenp = 0;
1635 return 0;
1636 }
1637
1638 data += *ppos;
1639 len -= *ppos;
1640
1641 if (len > *lenp)
1642 len = *lenp;
1643 if (len)
1644 if(copy_to_user(buffer, data, len))
1645 return -EFAULT;
1646 if (len < *lenp) {
1647 if(put_user('\n', ((char __user *) buffer) + len))
1648 return -EFAULT;
1649 len++;
1650 }
1651 *lenp = len;
1652 *ppos += len;
1653 }
1654 return 0;
1655 }
1656
1657 /**
1658 * proc_dostring - read a string sysctl
1659 * @table: the sysctl table
1660 * @write: %TRUE if this is a write to the sysctl file
1661 * @filp: the file structure
1662 * @buffer: the user buffer
1663 * @lenp: the size of the user buffer
1664 * @ppos: file position
1665 *
1666 * Reads/writes a string from/to the user buffer. If the kernel
1667 * buffer provided is not large enough to hold the string, the
1668 * string is truncated. The copied string is %NULL-terminated.
1669 * If the string is being read by the user process, it is copied
1670 * and a newline '\n' is added. It is truncated if the buffer is
1671 * not large enough.
1672 *
1673 * Returns 0 on success.
1674 */
1675 int proc_dostring(ctl_table *table, int write, struct file *filp,
1676 void __user *buffer, size_t *lenp, loff_t *ppos)
1677 {
1678 return _proc_do_string(table->data, table->maxlen, write, filp,
1679 buffer, lenp, ppos);
1680 }
1681
1682
1683 static int do_proc_dointvec_conv(int *negp, unsigned long *lvalp,
1684 int *valp,
1685 int write, void *data)
1686 {
1687 if (write) {
1688 *valp = *negp ? -*lvalp : *lvalp;
1689 } else {
1690 int val = *valp;
1691 if (val < 0) {
1692 *negp = -1;
1693 *lvalp = (unsigned long)-val;
1694 } else {
1695 *negp = 0;
1696 *lvalp = (unsigned long)val;
1697 }
1698 }
1699 return 0;
1700 }
1701
1702 static int __do_proc_dointvec(void *tbl_data, ctl_table *table,
1703 int write, struct file *filp, void __user *buffer,
1704 size_t *lenp, loff_t *ppos,
1705 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1706 int write, void *data),
1707 void *data)
1708 {
1709 #define TMPBUFLEN 21
1710 int *i, vleft, first=1, neg, val;
1711 unsigned long lval;
1712 size_t left, len;
1713
1714 char buf[TMPBUFLEN], *p;
1715 char __user *s = buffer;
1716
1717 if (!tbl_data || !table->maxlen || !*lenp ||
1718 (*ppos && !write)) {
1719 *lenp = 0;
1720 return 0;
1721 }
1722
1723 i = (int *) tbl_data;
1724 vleft = table->maxlen / sizeof(*i);
1725 left = *lenp;
1726
1727 if (!conv)
1728 conv = do_proc_dointvec_conv;
1729
1730 for (; left && vleft--; i++, first=0) {
1731 if (write) {
1732 while (left) {
1733 char c;
1734 if (get_user(c, s))
1735 return -EFAULT;
1736 if (!isspace(c))
1737 break;
1738 left--;
1739 s++;
1740 }
1741 if (!left)
1742 break;
1743 neg = 0;
1744 len = left;
1745 if (len > sizeof(buf) - 1)
1746 len = sizeof(buf) - 1;
1747 if (copy_from_user(buf, s, len))
1748 return -EFAULT;
1749 buf[len] = 0;
1750 p = buf;
1751 if (*p == '-' && left > 1) {
1752 neg = 1;
1753 p++;
1754 }
1755 if (*p < '0' || *p > '9')
1756 break;
1757
1758 lval = simple_strtoul(p, &p, 0);
1759
1760 len = p-buf;
1761 if ((len < left) && *p && !isspace(*p))
1762 break;
1763 if (neg)
1764 val = -val;
1765 s += len;
1766 left -= len;
1767
1768 if (conv(&neg, &lval, i, 1, data))
1769 break;
1770 } else {
1771 p = buf;
1772 if (!first)
1773 *p++ = '\t';
1774
1775 if (conv(&neg, &lval, i, 0, data))
1776 break;
1777
1778 sprintf(p, "%s%lu", neg ? "-" : "", lval);
1779 len = strlen(buf);
1780 if (len > left)
1781 len = left;
1782 if(copy_to_user(s, buf, len))
1783 return -EFAULT;
1784 left -= len;
1785 s += len;
1786 }
1787 }
1788
1789 if (!write && !first && left) {
1790 if(put_user('\n', s))
1791 return -EFAULT;
1792 left--, s++;
1793 }
1794 if (write) {
1795 while (left) {
1796 char c;
1797 if (get_user(c, s++))
1798 return -EFAULT;
1799 if (!isspace(c))
1800 break;
1801 left--;
1802 }
1803 }
1804 if (write && first)
1805 return -EINVAL;
1806 *lenp -= left;
1807 *ppos += *lenp;
1808 return 0;
1809 #undef TMPBUFLEN
1810 }
1811
1812 static int do_proc_dointvec(ctl_table *table, int write, struct file *filp,
1813 void __user *buffer, size_t *lenp, loff_t *ppos,
1814 int (*conv)(int *negp, unsigned long *lvalp, int *valp,
1815 int write, void *data),
1816 void *data)
1817 {
1818 return __do_proc_dointvec(table->data, table, write, filp,
1819 buffer, lenp, ppos, conv, data);
1820 }
1821
1822 /**
1823 * proc_dointvec - read a vector of integers
1824 * @table: the sysctl table
1825 * @write: %TRUE if this is a write to the sysctl file
1826 * @filp: the file structure
1827 * @buffer: the user buffer
1828 * @lenp: the size of the user buffer
1829 * @ppos: file position
1830 *
1831 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1832 * values from/to the user buffer, treated as an ASCII string.
1833 *
1834 * Returns 0 on success.
1835 */
1836 int proc_dointvec(ctl_table *table, int write, struct file *filp,
1837 void __user *buffer, size_t *lenp, loff_t *ppos)
1838 {
1839 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1840 NULL,NULL);
1841 }
1842
1843 #define OP_SET 0
1844 #define OP_AND 1
1845 #define OP_OR 2
1846
1847 static int do_proc_dointvec_bset_conv(int *negp, unsigned long *lvalp,
1848 int *valp,
1849 int write, void *data)
1850 {
1851 int op = *(int *)data;
1852 if (write) {
1853 int val = *negp ? -*lvalp : *lvalp;
1854 switch(op) {
1855 case OP_SET: *valp = val; break;
1856 case OP_AND: *valp &= val; break;
1857 case OP_OR: *valp |= val; break;
1858 }
1859 } else {
1860 int val = *valp;
1861 if (val < 0) {
1862 *negp = -1;
1863 *lvalp = (unsigned long)-val;
1864 } else {
1865 *negp = 0;
1866 *lvalp = (unsigned long)val;
1867 }
1868 }
1869 return 0;
1870 }
1871
1872 /*
1873 * init may raise the set.
1874 */
1875
1876 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
1877 void __user *buffer, size_t *lenp, loff_t *ppos)
1878 {
1879 int op;
1880
1881 if (write && !capable(CAP_SYS_MODULE)) {
1882 return -EPERM;
1883 }
1884
1885 op = is_init(current) ? OP_SET : OP_AND;
1886 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1887 do_proc_dointvec_bset_conv,&op);
1888 }
1889
1890 /*
1891 * Taint values can only be increased
1892 */
1893 static int proc_dointvec_taint(ctl_table *table, int write, struct file *filp,
1894 void __user *buffer, size_t *lenp, loff_t *ppos)
1895 {
1896 int op;
1897
1898 if (write && !capable(CAP_SYS_ADMIN))
1899 return -EPERM;
1900
1901 op = OP_OR;
1902 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
1903 do_proc_dointvec_bset_conv,&op);
1904 }
1905
1906 struct do_proc_dointvec_minmax_conv_param {
1907 int *min;
1908 int *max;
1909 };
1910
1911 static int do_proc_dointvec_minmax_conv(int *negp, unsigned long *lvalp,
1912 int *valp,
1913 int write, void *data)
1914 {
1915 struct do_proc_dointvec_minmax_conv_param *param = data;
1916 if (write) {
1917 int val = *negp ? -*lvalp : *lvalp;
1918 if ((param->min && *param->min > val) ||
1919 (param->max && *param->max < val))
1920 return -EINVAL;
1921 *valp = val;
1922 } else {
1923 int val = *valp;
1924 if (val < 0) {
1925 *negp = -1;
1926 *lvalp = (unsigned long)-val;
1927 } else {
1928 *negp = 0;
1929 *lvalp = (unsigned long)val;
1930 }
1931 }
1932 return 0;
1933 }
1934
1935 /**
1936 * proc_dointvec_minmax - read a vector of integers with min/max values
1937 * @table: the sysctl table
1938 * @write: %TRUE if this is a write to the sysctl file
1939 * @filp: the file structure
1940 * @buffer: the user buffer
1941 * @lenp: the size of the user buffer
1942 * @ppos: file position
1943 *
1944 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
1945 * values from/to the user buffer, treated as an ASCII string.
1946 *
1947 * This routine will ensure the values are within the range specified by
1948 * table->extra1 (min) and table->extra2 (max).
1949 *
1950 * Returns 0 on success.
1951 */
1952 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
1953 void __user *buffer, size_t *lenp, loff_t *ppos)
1954 {
1955 struct do_proc_dointvec_minmax_conv_param param = {
1956 .min = (int *) table->extra1,
1957 .max = (int *) table->extra2,
1958 };
1959 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
1960 do_proc_dointvec_minmax_conv, &param);
1961 }
1962
1963 static int __do_proc_doulongvec_minmax(void *data, ctl_table *table, int write,
1964 struct file *filp,
1965 void __user *buffer,
1966 size_t *lenp, loff_t *ppos,
1967 unsigned long convmul,
1968 unsigned long convdiv)
1969 {
1970 #define TMPBUFLEN 21
1971 unsigned long *i, *min, *max, val;
1972 int vleft, first=1, neg;
1973 size_t len, left;
1974 char buf[TMPBUFLEN], *p;
1975 char __user *s = buffer;
1976
1977 if (!data || !table->maxlen || !*lenp ||
1978 (*ppos && !write)) {
1979 *lenp = 0;
1980 return 0;
1981 }
1982
1983 i = (unsigned long *) data;
1984 min = (unsigned long *) table->extra1;
1985 max = (unsigned long *) table->extra2;
1986 vleft = table->maxlen / sizeof(unsigned long);
1987 left = *lenp;
1988
1989 for (; left && vleft--; i++, min++, max++, first=0) {
1990 if (write) {
1991 while (left) {
1992 char c;
1993 if (get_user(c, s))
1994 return -EFAULT;
1995 if (!isspace(c))
1996 break;
1997 left--;
1998 s++;
1999 }
2000 if (!left)
2001 break;
2002 neg = 0;
2003 len = left;
2004 if (len > TMPBUFLEN-1)
2005 len = TMPBUFLEN-1;
2006 if (copy_from_user(buf, s, len))
2007 return -EFAULT;
2008 buf[len] = 0;
2009 p = buf;
2010 if (*p == '-' && left > 1) {
2011 neg = 1;
2012 p++;
2013 }
2014 if (*p < '0' || *p > '9')
2015 break;
2016 val = simple_strtoul(p, &p, 0) * convmul / convdiv ;
2017 len = p-buf;
2018 if ((len < left) && *p && !isspace(*p))
2019 break;
2020 if (neg)
2021 val = -val;
2022 s += len;
2023 left -= len;
2024
2025 if(neg)
2026 continue;
2027 if ((min && val < *min) || (max && val > *max))
2028 continue;
2029 *i = val;
2030 } else {
2031 p = buf;
2032 if (!first)
2033 *p++ = '\t';
2034 sprintf(p, "%lu", convdiv * (*i) / convmul);
2035 len = strlen(buf);
2036 if (len > left)
2037 len = left;
2038 if(copy_to_user(s, buf, len))
2039 return -EFAULT;
2040 left -= len;
2041 s += len;
2042 }
2043 }
2044
2045 if (!write && !first && left) {
2046 if(put_user('\n', s))
2047 return -EFAULT;
2048 left--, s++;
2049 }
2050 if (write) {
2051 while (left) {
2052 char c;
2053 if (get_user(c, s++))
2054 return -EFAULT;
2055 if (!isspace(c))
2056 break;
2057 left--;
2058 }
2059 }
2060 if (write && first)
2061 return -EINVAL;
2062 *lenp -= left;
2063 *ppos += *lenp;
2064 return 0;
2065 #undef TMPBUFLEN
2066 }
2067
2068 static int do_proc_doulongvec_minmax(ctl_table *table, int write,
2069 struct file *filp,
2070 void __user *buffer,
2071 size_t *lenp, loff_t *ppos,
2072 unsigned long convmul,
2073 unsigned long convdiv)
2074 {
2075 return __do_proc_doulongvec_minmax(table->data, table, write,
2076 filp, buffer, lenp, ppos, convmul, convdiv);
2077 }
2078
2079 /**
2080 * proc_doulongvec_minmax - read a vector of long integers with min/max values
2081 * @table: the sysctl table
2082 * @write: %TRUE if this is a write to the sysctl file
2083 * @filp: the file structure
2084 * @buffer: the user buffer
2085 * @lenp: the size of the user buffer
2086 * @ppos: file position
2087 *
2088 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2089 * values from/to the user buffer, treated as an ASCII string.
2090 *
2091 * This routine will ensure the values are within the range specified by
2092 * table->extra1 (min) and table->extra2 (max).
2093 *
2094 * Returns 0 on success.
2095 */
2096 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2097 void __user *buffer, size_t *lenp, loff_t *ppos)
2098 {
2099 return do_proc_doulongvec_minmax(table, write, filp, buffer, lenp, ppos, 1l, 1l);
2100 }
2101
2102 /**
2103 * proc_doulongvec_ms_jiffies_minmax - read a vector of millisecond values with min/max values
2104 * @table: the sysctl table
2105 * @write: %TRUE if this is a write to the sysctl file
2106 * @filp: the file structure
2107 * @buffer: the user buffer
2108 * @lenp: the size of the user buffer
2109 * @ppos: file position
2110 *
2111 * Reads/writes up to table->maxlen/sizeof(unsigned long) unsigned long
2112 * values from/to the user buffer, treated as an ASCII string. The values
2113 * are treated as milliseconds, and converted to jiffies when they are stored.
2114 *
2115 * This routine will ensure the values are within the range specified by
2116 * table->extra1 (min) and table->extra2 (max).
2117 *
2118 * Returns 0 on success.
2119 */
2120 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2121 struct file *filp,
2122 void __user *buffer,
2123 size_t *lenp, loff_t *ppos)
2124 {
2125 return do_proc_doulongvec_minmax(table, write, filp, buffer,
2126 lenp, ppos, HZ, 1000l);
2127 }
2128
2129
2130 static int do_proc_dointvec_jiffies_conv(int *negp, unsigned long *lvalp,
2131 int *valp,
2132 int write, void *data)
2133 {
2134 if (write) {
2135 if (*lvalp > LONG_MAX / HZ)
2136 return 1;
2137 *valp = *negp ? -(*lvalp*HZ) : (*lvalp*HZ);
2138 } else {
2139 int val = *valp;
2140 unsigned long lval;
2141 if (val < 0) {
2142 *negp = -1;
2143 lval = (unsigned long)-val;
2144 } else {
2145 *negp = 0;
2146 lval = (unsigned long)val;
2147 }
2148 *lvalp = lval / HZ;
2149 }
2150 return 0;
2151 }
2152
2153 static int do_proc_dointvec_userhz_jiffies_conv(int *negp, unsigned long *lvalp,
2154 int *valp,
2155 int write, void *data)
2156 {
2157 if (write) {
2158 if (USER_HZ < HZ && *lvalp > (LONG_MAX / HZ) * USER_HZ)
2159 return 1;
2160 *valp = clock_t_to_jiffies(*negp ? -*lvalp : *lvalp);
2161 } else {
2162 int val = *valp;
2163 unsigned long lval;
2164 if (val < 0) {
2165 *negp = -1;
2166 lval = (unsigned long)-val;
2167 } else {
2168 *negp = 0;
2169 lval = (unsigned long)val;
2170 }
2171 *lvalp = jiffies_to_clock_t(lval);
2172 }
2173 return 0;
2174 }
2175
2176 static int do_proc_dointvec_ms_jiffies_conv(int *negp, unsigned long *lvalp,
2177 int *valp,
2178 int write, void *data)
2179 {
2180 if (write) {
2181 *valp = msecs_to_jiffies(*negp ? -*lvalp : *lvalp);
2182 } else {
2183 int val = *valp;
2184 unsigned long lval;
2185 if (val < 0) {
2186 *negp = -1;
2187 lval = (unsigned long)-val;
2188 } else {
2189 *negp = 0;
2190 lval = (unsigned long)val;
2191 }
2192 *lvalp = jiffies_to_msecs(lval);
2193 }
2194 return 0;
2195 }
2196
2197 /**
2198 * proc_dointvec_jiffies - read a vector of integers as seconds
2199 * @table: the sysctl table
2200 * @write: %TRUE if this is a write to the sysctl file
2201 * @filp: the file structure
2202 * @buffer: the user buffer
2203 * @lenp: the size of the user buffer
2204 * @ppos: file position
2205 *
2206 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2207 * values from/to the user buffer, treated as an ASCII string.
2208 * The values read are assumed to be in seconds, and are converted into
2209 * jiffies.
2210 *
2211 * Returns 0 on success.
2212 */
2213 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2214 void __user *buffer, size_t *lenp, loff_t *ppos)
2215 {
2216 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2217 do_proc_dointvec_jiffies_conv,NULL);
2218 }
2219
2220 /**
2221 * proc_dointvec_userhz_jiffies - read a vector of integers as 1/USER_HZ seconds
2222 * @table: the sysctl table
2223 * @write: %TRUE if this is a write to the sysctl file
2224 * @filp: the file structure
2225 * @buffer: the user buffer
2226 * @lenp: the size of the user buffer
2227 * @ppos: pointer to the file position
2228 *
2229 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2230 * values from/to the user buffer, treated as an ASCII string.
2231 * The values read are assumed to be in 1/USER_HZ seconds, and
2232 * are converted into jiffies.
2233 *
2234 * Returns 0 on success.
2235 */
2236 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2237 void __user *buffer, size_t *lenp, loff_t *ppos)
2238 {
2239 return do_proc_dointvec(table,write,filp,buffer,lenp,ppos,
2240 do_proc_dointvec_userhz_jiffies_conv,NULL);
2241 }
2242
2243 /**
2244 * proc_dointvec_ms_jiffies - read a vector of integers as 1 milliseconds
2245 * @table: the sysctl table
2246 * @write: %TRUE if this is a write to the sysctl file
2247 * @filp: the file structure
2248 * @buffer: the user buffer
2249 * @lenp: the size of the user buffer
2250 * @ppos: file position
2251 * @ppos: the current position in the file
2252 *
2253 * Reads/writes up to table->maxlen/sizeof(unsigned int) integer
2254 * values from/to the user buffer, treated as an ASCII string.
2255 * The values read are assumed to be in 1/1000 seconds, and
2256 * are converted into jiffies.
2257 *
2258 * Returns 0 on success.
2259 */
2260 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2261 void __user *buffer, size_t *lenp, loff_t *ppos)
2262 {
2263 return do_proc_dointvec(table, write, filp, buffer, lenp, ppos,
2264 do_proc_dointvec_ms_jiffies_conv, NULL);
2265 }
2266
2267 static int proc_do_cad_pid(ctl_table *table, int write, struct file *filp,
2268 void __user *buffer, size_t *lenp, loff_t *ppos)
2269 {
2270 struct pid *new_pid;
2271 pid_t tmp;
2272 int r;
2273
2274 tmp = pid_nr(cad_pid);
2275
2276 r = __do_proc_dointvec(&tmp, table, write, filp, buffer,
2277 lenp, ppos, NULL, NULL);
2278 if (r || !write)
2279 return r;
2280
2281 new_pid = find_get_pid(tmp);
2282 if (!new_pid)
2283 return -ESRCH;
2284
2285 put_pid(xchg(&cad_pid, new_pid));
2286 return 0;
2287 }
2288
2289 #else /* CONFIG_PROC_FS */
2290
2291 int proc_dostring(ctl_table *table, int write, struct file *filp,
2292 void __user *buffer, size_t *lenp, loff_t *ppos)
2293 {
2294 return -ENOSYS;
2295 }
2296
2297 int proc_dointvec(ctl_table *table, int write, struct file *filp,
2298 void __user *buffer, size_t *lenp, loff_t *ppos)
2299 {
2300 return -ENOSYS;
2301 }
2302
2303 int proc_dointvec_bset(ctl_table *table, int write, struct file *filp,
2304 void __user *buffer, size_t *lenp, loff_t *ppos)
2305 {
2306 return -ENOSYS;
2307 }
2308
2309 int proc_dointvec_minmax(ctl_table *table, int write, struct file *filp,
2310 void __user *buffer, size_t *lenp, loff_t *ppos)
2311 {
2312 return -ENOSYS;
2313 }
2314
2315 int proc_dointvec_jiffies(ctl_table *table, int write, struct file *filp,
2316 void __user *buffer, size_t *lenp, loff_t *ppos)
2317 {
2318 return -ENOSYS;
2319 }
2320
2321 int proc_dointvec_userhz_jiffies(ctl_table *table, int write, struct file *filp,
2322 void __user *buffer, size_t *lenp, loff_t *ppos)
2323 {
2324 return -ENOSYS;
2325 }
2326
2327 int proc_dointvec_ms_jiffies(ctl_table *table, int write, struct file *filp,
2328 void __user *buffer, size_t *lenp, loff_t *ppos)
2329 {
2330 return -ENOSYS;
2331 }
2332
2333 int proc_doulongvec_minmax(ctl_table *table, int write, struct file *filp,
2334 void __user *buffer, size_t *lenp, loff_t *ppos)
2335 {
2336 return -ENOSYS;
2337 }
2338
2339 int proc_doulongvec_ms_jiffies_minmax(ctl_table *table, int write,
2340 struct file *filp,
2341 void __user *buffer,
2342 size_t *lenp, loff_t *ppos)
2343 {
2344 return -ENOSYS;
2345 }
2346
2347
2348 #endif /* CONFIG_PROC_FS */
2349
2350
2351 #ifdef CONFIG_SYSCTL_SYSCALL
2352 /*
2353 * General sysctl support routines
2354 */
2355
2356 /* The generic string strategy routine: */
2357 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2358 void __user *oldval, size_t __user *oldlenp,
2359 void __user *newval, size_t newlen)
2360 {
2361 if (!table->data || !table->maxlen)
2362 return -ENOTDIR;
2363
2364 if (oldval && oldlenp) {
2365 size_t bufsize;
2366 if (get_user(bufsize, oldlenp))
2367 return -EFAULT;
2368 if (bufsize) {
2369 size_t len = strlen(table->data), copied;
2370
2371 /* This shouldn't trigger for a well-formed sysctl */
2372 if (len > table->maxlen)
2373 len = table->maxlen;
2374
2375 /* Copy up to a max of bufsize-1 bytes of the string */
2376 copied = (len >= bufsize) ? bufsize - 1 : len;
2377
2378 if (copy_to_user(oldval, table->data, copied) ||
2379 put_user(0, (char __user *)(oldval + copied)))
2380 return -EFAULT;
2381 if (put_user(len, oldlenp))
2382 return -EFAULT;
2383 }
2384 }
2385 if (newval && newlen) {
2386 size_t len = newlen;
2387 if (len > table->maxlen)
2388 len = table->maxlen;
2389 if(copy_from_user(table->data, newval, len))
2390 return -EFAULT;
2391 if (len == table->maxlen)
2392 len--;
2393 ((char *) table->data)[len] = 0;
2394 }
2395 return 1;
2396 }
2397
2398 /*
2399 * This function makes sure that all of the integers in the vector
2400 * are between the minimum and maximum values given in the arrays
2401 * table->extra1 and table->extra2, respectively.
2402 */
2403 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2404 void __user *oldval, size_t __user *oldlenp,
2405 void __user *newval, size_t newlen)
2406 {
2407
2408 if (newval && newlen) {
2409 int __user *vec = (int __user *) newval;
2410 int *min = (int *) table->extra1;
2411 int *max = (int *) table->extra2;
2412 size_t length;
2413 int i;
2414
2415 if (newlen % sizeof(int) != 0)
2416 return -EINVAL;
2417
2418 if (!table->extra1 && !table->extra2)
2419 return 0;
2420
2421 if (newlen > table->maxlen)
2422 newlen = table->maxlen;
2423 length = newlen / sizeof(int);
2424
2425 for (i = 0; i < length; i++) {
2426 int value;
2427 if (get_user(value, vec + i))
2428 return -EFAULT;
2429 if (min && value < min[i])
2430 return -EINVAL;
2431 if (max && value > max[i])
2432 return -EINVAL;
2433 }
2434 }
2435 return 0;
2436 }
2437
2438 /* Strategy function to convert jiffies to seconds */
2439 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2440 void __user *oldval, size_t __user *oldlenp,
2441 void __user *newval, size_t newlen)
2442 {
2443 if (oldval && oldlenp) {
2444 size_t olen;
2445
2446 if (get_user(olen, oldlenp))
2447 return -EFAULT;
2448 if (olen) {
2449 int val;
2450
2451 if (olen < sizeof(int))
2452 return -EINVAL;
2453
2454 val = *(int *)(table->data) / HZ;
2455 if (put_user(val, (int __user *)oldval))
2456 return -EFAULT;
2457 if (put_user(sizeof(int), oldlenp))
2458 return -EFAULT;
2459 }
2460 }
2461 if (newval && newlen) {
2462 int new;
2463 if (newlen != sizeof(int))
2464 return -EINVAL;
2465 if (get_user(new, (int __user *)newval))
2466 return -EFAULT;
2467 *(int *)(table->data) = new*HZ;
2468 }
2469 return 1;
2470 }
2471
2472 /* Strategy function to convert jiffies to seconds */
2473 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2474 void __user *oldval, size_t __user *oldlenp,
2475 void __user *newval, size_t newlen)
2476 {
2477 if (oldval && oldlenp) {
2478 size_t olen;
2479
2480 if (get_user(olen, oldlenp))
2481 return -EFAULT;
2482 if (olen) {
2483 int val;
2484
2485 if (olen < sizeof(int))
2486 return -EINVAL;
2487
2488 val = jiffies_to_msecs(*(int *)(table->data));
2489 if (put_user(val, (int __user *)oldval))
2490 return -EFAULT;
2491 if (put_user(sizeof(int), oldlenp))
2492 return -EFAULT;
2493 }
2494 }
2495 if (newval && newlen) {
2496 int new;
2497 if (newlen != sizeof(int))
2498 return -EINVAL;
2499 if (get_user(new, (int __user *)newval))
2500 return -EFAULT;
2501 *(int *)(table->data) = msecs_to_jiffies(new);
2502 }
2503 return 1;
2504 }
2505
2506
2507
2508 #else /* CONFIG_SYSCTL_SYSCALL */
2509
2510
2511 asmlinkage long sys_sysctl(struct __sysctl_args __user *args)
2512 {
2513 static int msg_count;
2514 struct __sysctl_args tmp;
2515 int name[CTL_MAXNAME];
2516 int i;
2517
2518 /* Read in the sysctl name for better debug message logging */
2519 if (copy_from_user(&tmp, args, sizeof(tmp)))
2520 return -EFAULT;
2521 if (tmp.nlen <= 0 || tmp.nlen >= CTL_MAXNAME)
2522 return -ENOTDIR;
2523 for (i = 0; i < tmp.nlen; i++)
2524 if (get_user(name[i], tmp.name + i))
2525 return -EFAULT;
2526
2527 /* Ignore accesses to kernel.version */
2528 if ((tmp.nlen == 2) && (name[0] == CTL_KERN) && (name[1] == KERN_VERSION))
2529 goto out;
2530
2531 if (msg_count < 5) {
2532 msg_count++;
2533 printk(KERN_INFO
2534 "warning: process `%s' used the removed sysctl "
2535 "system call with ", current->comm);
2536 for (i = 0; i < tmp.nlen; i++)
2537 printk("%d.", name[i]);
2538 printk("\n");
2539 }
2540 out:
2541 return -ENOSYS;
2542 }
2543
2544 int sysctl_string(ctl_table *table, int __user *name, int nlen,
2545 void __user *oldval, size_t __user *oldlenp,
2546 void __user *newval, size_t newlen)
2547 {
2548 return -ENOSYS;
2549 }
2550
2551 int sysctl_intvec(ctl_table *table, int __user *name, int nlen,
2552 void __user *oldval, size_t __user *oldlenp,
2553 void __user *newval, size_t newlen)
2554 {
2555 return -ENOSYS;
2556 }
2557
2558 int sysctl_jiffies(ctl_table *table, int __user *name, int nlen,
2559 void __user *oldval, size_t __user *oldlenp,
2560 void __user *newval, size_t newlen)
2561 {
2562 return -ENOSYS;
2563 }
2564
2565 int sysctl_ms_jiffies(ctl_table *table, int __user *name, int nlen,
2566 void __user *oldval, size_t __user *oldlenp,
2567 void __user *newval, size_t newlen)
2568 {
2569 return -ENOSYS;
2570 }
2571
2572 #endif /* CONFIG_SYSCTL_SYSCALL */
2573
2574 /*
2575 * No sense putting this after each symbol definition, twice,
2576 * exception granted :-)
2577 */
2578 EXPORT_SYMBOL(proc_dointvec);
2579 EXPORT_SYMBOL(proc_dointvec_jiffies);
2580 EXPORT_SYMBOL(proc_dointvec_minmax);
2581 EXPORT_SYMBOL(proc_dointvec_userhz_jiffies);
2582 EXPORT_SYMBOL(proc_dointvec_ms_jiffies);
2583 EXPORT_SYMBOL(proc_dostring);
2584 EXPORT_SYMBOL(proc_doulongvec_minmax);
2585 EXPORT_SYMBOL(proc_doulongvec_ms_jiffies_minmax);
2586 EXPORT_SYMBOL(register_sysctl_table);
2587 EXPORT_SYMBOL(sysctl_intvec);
2588 EXPORT_SYMBOL(sysctl_jiffies);
2589 EXPORT_SYMBOL(sysctl_ms_jiffies);
2590 EXPORT_SYMBOL(sysctl_string);
2591 EXPORT_SYMBOL(unregister_sysctl_table);
Ubuntu Artful kernel mirror